[tex-implementors] Re: [tex-live] LM as the default outline font?

Philip TAYLOR P.Taylor at Rhul.Ac.Uk
Tue Mar 29 18:11:24 CEST 2005

Frank Küster wrote:


> If I know that there will never be a lmodern version 0.100 (and 0.101,
> 0.299, 0.903, 0.993), then there is nothing wrong with it. 

There may well have /been/ versions 0.100 .. 0.903, but these
will all ne of only archival interest, since there is now
a version 0.982

>  But this only works for the LM fonts.  Anybody who wants to describe and
> implement a scheme for version comparison that is generally applicable
> to different pieces of software in a collection (like CTAN) or
> distribution (like any Linux distribution), he will have to arrange for
> version numbers in the three-digit range.  

Why, Frank ?  Where is it written on tablets of stone (or elsewhere)
that versions must be differentiable by the first three significant
digits.  Why can Staszek et al not propose LM-fonts V0.9821, being a
subtle improvement on V0.982, then V0.99, 0.991, 0.9911, 0.9912,
0.99121, and so on ?  If Knuth may elect to number TeX as

	V3.14159265359879663 ...

and MF likewise (for e, rather than for pi) then why may others
not follow his lead of using real numbers to differentiate versions ?

> Of course my less version 382
> must be higher than some ancient version 39, 

Yes, if "less" is numbered using positive integers

> and of course libconvert-binhex-perl version 1.119 must be higher than 1.12.  

"must" ?  I honestly fail to see any logic in this last statement at
all.  /Why/ "must" libconvert-binhex-perl version 1.119 be higher than 1.12 ?
It would certainly not appear to be so, if by "higher" you imply "more recent".

> There is no way to tell that lmodern will have 1.00 after 0.99 without knowing
> about lmodern, 

Yes, I agree : unless one knows how packages are numbered, one has
no way of differentiating between earlier and later versions based
solely on their name.  But we /do/ know about the numbering scheme
for LM Modern, and therefore there is no problem.  There /does/
seem to be a problem for "libconvert-binhex-perl", if version 1.119
is really "higher" (more recent) than 1.12, but that would appear
to be a problem for the author(s)/creator(s)/maintainer(s) of
"libconvert-binhex-perl" rather than for Staszek et al.

> in other words it is impossible in a generally applicable
> version comparison.

Not sure what this statement means ...

> And if 0.100 is higher than 0.99, then 0.982 will  be higher, too.

and if the Holy Father were a Protestant, then the entire Vatican
staff would be out of a job!  But he isn't, any more than
0.100 is higher than 0.99, so what are we to gain by discussing
counter-factuals ?

** Phil.

More information about the tex-live mailing list